New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
1.7 allocates where 1.6 doesn't #41488
Comments
What is |
Would also be helpful to bisect if you can. |
Good point. I think I can narrow it down a little. I'll edit the OP in a minute... |
I've never thought about bisecting such a big codebase. I assume each step takes a while? Are there utilities to help with this? Or maybe it's just a little one-off script each time? |
Not sure this is what you asked for: git-bisect can help, e.g.: https://git-scm.com/docs/git-bisect |
Thanks, I've used |
This is interesting: |
Also code_llvm confirms there are no heap allocations in the generated code. |
Oh, that's very weird. It's strange that 1.6 consistently gives 0 and 1.7 gives 32, but maybe it's a non-issue, and I should just change it to |
We changed the |
Good to know, thanks @vtjnash |
In KeywordCalls.jl we have some tests like this:
The
@kwstruct Foo(a,b)
expands towhere
_sort
is like this:The
aliased
bit doesn't really apply here (no aliases, so it defaults toidentity
. But just in case, it'sThese are passing in 1.6 but failing in the
nightly
build, and also failing in 1.7-beta. Some details here.Here's the result in my terminal for 1.7-beta:
Also...
Also... @simeonschaub you're familiar with this code, would you have time to have a look?
The text was updated successfully, but these errors were encountered: